Pinned Tweet
We built a multi-agent harness that runs all your agents, context, and loops
> Turned it into an everything app for my life because why not
> One app runs my portfolio, calendar, inbox, health, and more
Comment below to join the waitlist
time for a new slogan @Railway
An urgent notice from the Railway team
blog.railway.com/p/security-…
if your github looks anything like this, drop what you’re building
mine’s mostly apps and a personal agent harness lately
This is what happens when a team really find product market fit
solana:METvsvVRapdj9cFLzq4Tr43xK4tAjQfwX76z3n6mWQL
Great work @MeteoraAG and @met_lparmy
Fun to see people are still using polyskill.ai even though I haven't posted about it in a long time
All organic usage
Not much but cool to see!
I'm generally not a p(doom) AI guy, but
A) it's alarming to see safety researchers at the frontier labs quitting and going public
B) the fact that they can't even sandbox their own agents in testing is absolutely ridiculous, e.g. the Hugging Face and Australian Government hacks
C) anyone claiming a 0% chance of AI killing humanity isn't being serious
Europe have been heading down a dark path for a while now
Feels like its reached the point of no return given the leadership misaligned interest with the native population
Sometimes its better to leave the burning ship than move the mountain
It's easy to love open source models when you can get near frontier models at 10x less cost
Also love how @OpenRouter gets you an even better deal both price and privacy-wise than the model developer.
Why does it feel like models are getting dumber
Sonnet 3.5 was a breakthrough
Opus 4.5 was elite cutting edge
But after Fabel using sonnet 5 or opus 4.8 for complex tasks just feels retarded
The hedonic treadmill is real
























